Chapter 6: Quadratic Forms

Definition 1. Let A be a symmetric matrix of order n, and let X =

(x1, x2, . . . , xn)T ∈ Rn. The function Q : Rn → R defined by

Q(X) = X T AX

is called a quadratic form in the n variables x1, x2, · · · , xn. The

matrix A is called the matrix of the quadratic form Q.
Remark 1. The quadratic form Q may also be defined by

X
n X
n
Q(X) = Q((x1, x2, · · · , xn)) = aij xixj
i=1 j=1

Chapter 6: Quadratic Forms 1

  
3 2
Example 1. Consider the symmetric matrix A = and let X =
2 5
(x1, x2)T . Then the quadratic form defined by A is

   
3 2 x1
Q(X) = (x1, x2)
2 5 x2

= 3x21 + 5x22 + 2x1x2 + 2x2x1 = 3x21 + 5x22 + 4x1x2

Example 2. If a quadratic form Q is defined by Q(X) = Q((x, y, z)) =

3x2 − 8xy + 6xz + 10yz + y 2 − 3z 2, then we have

a11 = 3 a12 = a21 = −8/2 = −4 a13 = a31 = 6/2 = 3

a22 = 1 a23 = a32 = 10/2 = 5 a33 = −3

Chapter 6: Quadratic Forms 2

so that the matrix of the quadratic form is
 
3 −4 3
A =  −4 1 5 
3 5 −3

Definition 2.

a. Let A and B be square matrices of order n. We say that the two

matrices are congruent if there exists a nonsingular matrix P such that
B = P T AP .

b. Let Q and Q′ be two quadratic forms with respective matrices A and B,

respectively. We say that the two quadratic forms are equivalent if the
matrices A and B are congruent.

Chapter 6: Quadratic Forms 3

Example 3. Let Q be the quadratic form defined
 by Q(X) = Q((x, y)) =
2 1
2 2
2x + 2xy + 2y and with matrix A = and let Q′ be the quadratic
1 2  
3 0
form defined by Q′(X) = Q′((x, y)) = 3x2 + y 2, with matrix B = .
" # 0 1
√1
2
− √2
1
If we let P = √1 √1
, then we have
2 2

" #   " √1 #
√1 √1 2 1 − √12
P T AP = 2 2 2
− √12 √1
2
1 2 √1
2
√1
2
 
3 0
= =B
0 1

Since |P | = 1 ̸= 0, we know that P is nonsingular. Since B = P T AP ,

we see that A and B are congruent, so that their corresponding quadratic
forms Q and Q′ are equivalent.

Chapter 6: Quadratic Forms 4

 The following result shows that any quadratic form Q is equivalent to
a quadratic form Q′ whose coefficients are the eigenvalues of the matrix of
Q.
Theorem 1. Let Q be a quadratic form in n variables defined by Q(X) =
X T AX whose matrix is A. Then Q is equivalent to a quadratic form Q′
defined by

Q′(Y ) = Y T BY = λ1y12 + λ2y22 + · · · + λnyn2

where λ1, λ2, · · · , λn are the eigenvalues of A. Moreover, there exists

an orthogonal matrix P such that B = P T AP .

Proof. Since A is a symmetric matrix, we know that it can be diagonalized

by an orthogonal matrix P . This means that the matrix

B = P −1AP

Chapter 6: Quadratic Forms 5

is a diagonal matrix whose diagonal entries are the eigenvalues of A, say
λ1, λ2, · · · , λn. Since P is orthogonal, it follows that P −1 = P T , and so
we have
B = P T AP
Since B is symmetric, it represents a quadratic form Q′ of n variables
defined by

Q′(Y ) = Y T BY = λ1y12 + λ2y22 + · · · + λnyn2

Since A and B are congruent, it is clear that Q and Q′ are equivalent

quadratic forms.

Example 4. Consider the quadratic form Q defined by

Q(X) = Q((x1, x2, x3)) = −x21 + x22 + x23 + 2x2x3

Chapter 6: Quadratic Forms 6

The matrix of this quadratic form is
 
−1 0 0
A =  0 1 1
0 1 1

It can be verified that this matrix has eigenvalues -1, 2 and 0. Thus, A is
congruent to the matrix
 
−1 0 0
B =  0 2 0
0 0 0

which is the matrix of the quadratic form Q′ defined by

Q′(Y ) = Q′((y1, y2, y3)) = −y12 + 2y22

Then Q and Q′ are equivalent quadratic forms.

Chapter 6: Quadratic Forms 7

Definition 3.

a. Let A be a symmetric matrix of order n. Then A is said to be positive

definite if X T AX > 0 for all X ∈ Rn\{0}.

b. A quadratic form Q is positive definite if its matrix is positive definite.

The following theorem gives a necessary and sufficient condition for a

matrix (and hence a quadratic form) to be positive definite.
Theorem 2. A symmetric matrix A is positive definite if and only if all
the eigenvalues of A are positive.

Chapter 6: Quadratic Forms 8

  
2 1 1
Example 5. Let A = 1 2 1. The characteristic polynomial of A is
1 1 2
given by
 
 x−2 −1 −1 

|xI3 − A| =  −1 x − 2 −1 
 −1 −1 x − 2 
= (x − 2)[(x − 2)2 − 1] − (−1)[(−1)(x − 2) − (−1)2]
+ (−1)[(−1)2 − (−1)(x − 2)]
= (x − 2)3 − (x − 2) + (−2)[1 + (x − 2)]
= x3 − 6x2 + +2 − x − 8 − 2x + 2
= x3 − 6x2 + 9x − 4 = (x − 1)2(x − 4)

which shows that the eigenvalues of A are all positive, namely 1, 1 and 4.
By Theorem 2, it follows that the given matrix is positive definite. The

Chapter 6: Quadratic Forms 9

quadratic form Q represented by A is defined by

Q(X) = Q((x, y, z)) = 2x2 + 2y 2 + 2z 2 + 2xy + 2xz + 2yz

= (x + y)2 + (x + z)2 + (y + z)2 > 0

is also positive definite.

Another characterization for positive definite matrices concerns what we

call leading principal minors of the matrix.
Definition 4. Let A = [aij ] be a square matrix of order n. The leading
principal minors of A are the determinants
 
 a11 a12 · · · a1k 
 
 a21 a22 · · · a2k 

|Pk | =   , k = 1, 2, . . . , n
.. .. · · · .. 
 
 ak1 ak2 · · · akk 

Chapter 6: Quadratic Forms 10

  
1 2 0
Example 6. Let A = −1 1 1 . Then the leading principal minors of
0 2 −1
A are

|P1| = |1| = 1
 
 1 2 
|P2| =   = (1)(1) − (−1)(2) = 3
−1 1 
 
 1 2 0 

|P3| = |A| =  −1 1 1  = −5
 0 2 −1 

The characterization of positive definite matrices in terms of leading

principal minors is given in the following result.

Chapter 6: Quadratic Forms 11

Theorem 3. Let A be a symmetric matrix of order n. A necessary and
sufficient condition for A to be positive definite is for all the leading
principal minors of A to be positive.
   
2 1 x
Example 7. Consider the matrix A = . For any X = ∈ R2\{0},
1 2 y
we have

X T AX = 2x2 + 2y 2 + 2xy = (x + y)2 + x2 + y 2 > 0

which shows that A is positive definite. The leading principal minors of A

are

|P1| = |2| = 2 > 0 and |P2| = |A| = 2(2) − 1(1) = 3 > 0

which shows that the conditions of the above theorem are satisfied.

Chapter 6: Quadratic Forms 12

Exercises

1. Write each quadratic form as X T AX where A is a symmetric matrix.

(a) −3x2 + 5xy − 2y 2
(b) 2x21 + 3x1x2 − 5x1x3 + 7x2x3
(c) 3x21 + x22 − 2x23 + x1x2 − x1x3 − 4x2x3
(d) x21 − 3x22 + 4x23 − 4x1x2 + 6x2x3

2. For each given symmetric matrix A, find a diagonal matrix D Ihal is

congruent to A.
   
−1 0 1 0 2 2
(a)  0 1 1 (c) 2 0 2
0 1 1 2 2 0
   
1 1 1 3 4 0
(b) 1 1 1 (d) 4 −3 0
1 1 1 0 0 5
Chapter 6: Quadratic Forms 13
3. Find a quadratic form of the type in Theorem 1 that is equivalent to the
given quadratic form.

(a) 2x2 + 4xy + 2y 2 (c) 2x21 + 4x22 + 4x23 + 10x2x3

(b) x21 + x22 + x23 + 2x1x2 (d) 2x21 + 3x22 + 3x23 + 4x2x3

4. Which of the given matrices are positive definite?

   
2 −1 1 0 0
(a)
−1 2 (c) 0 2 0 
  0 0 −3
3 1 0  
(b) 1 3 0 7 4 4
0 0 3 (d) 4 7 4
4 4 7

Chapter 6: Quadratic Forms 14